IBM Guest Lecture: Virtualise Everything and Save the Planet!

IT resource 'sprawl' is everywhere, servers are procured for separate
projects and under-utilised, the end effect is an environment that
consumes administration time, much 'wattage' and generates a lot of hot
air! The industry response is a call to 'consolidation', that is, to
gather many 'small' workloads into larger servers, increasing
utilisation and thus improving cost-efficiency, all in all a good
approach but this requires excellent negotiating skills to arrange
maintenance 'windows' with seventy different users -a difficult task at the best of times-
however, near-zero downtime systems such as the 'mainframe' can help
remove this problem, but within the 'UNIX' and Linux world, this was a
difficult problem to resolve. This lecture discusses the technologies
taken from the mainframe environment that allow UNIX servers to support
the operations required for large scale consolidation to reduce server
sprawl, power costs, etc by providing the ability to move running
workloads between separate physical servers without any 'service'
interruption.The discussion covers the technologies required for this
capability and their operation, including: The POWER hypervisor
virtualisation layer, virtual processors, virtual I/O (network and
disk), live partition mobility, application mobility and operating
system requirements.

The presenter is Peter Nutt, who joined IBM in 2000 and is based in the
UK. Peter primarily works on AIX systems and is a member of the STG
implementation team, he has expertise in the fields of virtualisation,
performance analysis, Oracle solutions and implementation. Peter has
co-authored three IBM Redbooks: 'Live Partition Mobility', 'Porting to
AIX 5L', 'A Practical Guide to RMC' and regularly presents at the EMEA
and Americas IBM Technical University conferences.